Autism detection based onautism spectrum quotientusing weightedaverage ensemblemethod

Abstract

Autism spectrum disorder (ASD) is a condition that occurs in an individual, wherein  it  is  accompanied  by  various  symptoms  such  as  difficulties  in socializing  with  others.  Early  detection  of  ASD  patients  can  assist  in preventing various symptoms caused by ASD. The focus of this research is to automate the diagnosis of ASD in an individual based on the results of the autism  spectrum  quotient  (AQ)  using  weighted  average  ensemble  method. Initially,  preprocessing  is  carried  out  on  the  dataset  to  ensure  optimal performance of the resulting model. In the preprocessing step, the filling of missing  values  and  feature  selection  occurs,  where  the  feature  selection method  being  utilized  is  p-value.  The  model  in  this  research  uses  the weighted average ensemble method, which is the model that combines three machine  learning  classification  algorithms.  Eight  classification  algorithms are tested to identify the three algorithms with the best performance, namely gaussian Naïve  Bayes(NB), logistic  regression  (LR),  and random  forest (RF).  Following  the  testing,  the  model  constructedusing  the  weighted average ensemble method exhibits the highest performance compared to the model built using a single classification algorithm. The performance matrix used   to   measure   the model’s  performance   is area   under   the   curve (AUC)/receiver  operating  characteristic  (ROC),  with  the  developed  model achievingan AUC/ROC value of 0.912.